CN116165885B - Model-free adaptive robust control method and system for high-speed train - Google Patents

Abstract

The invention relates to a model-free adaptive robust control method and system for high-speed trains, and relates to the field of automatic driving control of high-speed trains. The method includes: obtaining an improved Kalman filter; obtaining optimal estimates of control input and output at the previous moment. value and determine the output prediction value at the current moment; obtain the output measurement value of the high-speed train at the current moment; obtain the optimal estimate of the output at the current moment according to the gain and the output measurement value at the current moment; determine the pseudo-partial derivative; obtain the expected output signal; according The deviation is determined by the output optimal estimate and the expected output signal at the current moment; the high-speed train control signal at the current moment is determined based on the pseudo-partial derivative, deviation and the control input at the previous moment. The invention realizes the suppression of measurement disturbance, has better applicability, can obtain smaller tracking error and larger data signal-to-noise ratio, and improves the response speed of the controller.

Description

A model-free adaptive robust control method and system for high-speed trains

Technical field

The invention relates to the field of automatic driving control of high-speed trains, and in particular to a model-free adaptive robust control method of high-speed trains.

Background technique

Since the operation process of high-speed trains is a nonlinear dynamic system with complex and changeable environments and frequent changes in working conditions, it is extremely challenging to achieve high-precision tracking of train speed and displacement. Therefore, it is of great significance to consider the actual operating conditions of the train and design an anti-interference tracking control algorithm to achieve high-precision speed and displacement tracking control.

In order to solve the problem of disturbance suppression in train operation control, the adaptive output feedback trajectory tracking control method of high-speed trains has been proposed. The problems such as immeasurable speed, model parameter disturbance and unknown external disturbance in train operation are equivalent to nonlinear terms. Then compensate; the NLADRC (Nonlinear Active Distance Recept Controller) algorithm regards part of the mechanical delay of the system as internal disturbance, which reduces the dependence of the algorithm on the train model itself and solves the problem of difficulty in adjusting the parameters of the traditional algorithm. However, the above control strategy During design and stability analysis, it is often necessary to obtain system model parameters in advance, or linear approximation of the nonlinear part of the system is required. If the system model is unknown or there is a large disturbance, these methods are difficult to apply.

Regarding the measurement disturbance suppression problem of the model-free adaptive control (MFAC) method, the existing methods can be divided into two categories. The first category of methods is considered from the perspective of improving the MFAC control law: by introducing dead cells into the MFAC control law. The area link and attenuation factor make the control algorithm stop updating under certain circumstances to suppress measurement disturbances. This type of method has a better disturbance suppression effect when tracking a constant reference signal, but the control effect of this method needs to be improved when the reference signal changes time, and this method has the problem that the error threshold is difficult to select; the second type of method This is considered from the perspective of filtering noise data: the filtering ability of the tracking differentiator is used to modularize it with the MFAC controller to suppress measurement disturbances. However, the introduction of the tracking differentiator will cause phase loss in the signal, affecting the response speed of the controller, and the effect still needs to be improved.

Contents of the invention

The purpose of the present invention is to provide a model-free adaptive robust control method and system for high-speed trains. By deriving an improved Kalman filter (IKF), it can be used in situations where the controlled system model is unknown. Next, filter the noisy output data. Based on this IKF, a disturbance suppression MFAC scheme based on an improved Kalman filter is further designed to solve the problems existing in the background technology.

In order to achieve the above objects, the present invention provides the following solutions:

A model-free adaptive robust control method for high-speed trains, including:

Get a modified Kalman filter;

Based on the high-speed train control system, obtain the control input at the previous moment and the optimal estimate of the output at the previous moment;

Determine the output prediction value at the current moment based on the control input at the previous moment and the output optimal estimate at the previous moment;

Obtain the output measurement value of the high-speed train at the current moment;

Calculate the gain according to the modified Kalman filter;

Correct the output prediction value at the current moment according to the gain and the output measurement value at the current moment to obtain the optimal output estimate of the system at the current moment;

Determine the pseudo-partial derivative according to the control input at the previous moment, the output optimal estimate at the previous moment, and the output optimal estimate at the current moment;

Get the desired output signal;

Determine the deviation according to the output optimal estimate and the expected output signal at the current moment;

The high-speed train control signal at the current moment is determined based on the pseudo-partial derivative, deviation and the control input at the previous moment.

Optionally, the improved Kalman filter uses a dynamic linearized data model to describe the dynamic characteristics of the system.

Optionally, the improved Kalman filter is:

One-step prediction of output data:

One-step forecast error variance:

P(k|k-1)＝P(k-1|k-1)+X

Calculate the IKF gain:

Output the best estimate:

Update the estimated error variance:

P(k|k)＝[1-K(k)]P(k|k-1)

in, Output the best estimate for the previous moment/> one-step prediction,/> Output the best estimate for the current moment;/> is the pseudo partial derivative; U _L (k)=[u(k),...,u(k-L+1)] ^T , u(k) is the control variable, L is the linearization length; P(k-1| k-1) is the estimated error variance at the previous moment, P(k|k-1) is the one-step prediction error variance, P(k|k) is the estimated error variance at the current moment; K(k) is the IKF gain; v _m ( k)=v(k)+f(k) is the output measurement value of the system when it is disturbed, v(k) is the actual output of the system, f(k) is the measurement interference; X is the noise variance.

Optionally, use the following steps to determine the estimated value of the pseudo-partial derivative:

Introducing the parameter estimation criterion function:

Derivating the above equation with respect to Φ _P,L (k) and setting it to zero, the following pseudo-partial derivative estimation algorithm is obtained:

In order to enhance the robustness of the parameter estimation algorithm, a parameter reset algorithm is given:

like or |ΔU _L (k-1)|≤ε or/>

but

in, is the pseudo partial derivative; U _L (k)=[u(k),...,u(k-L+1)] ^T , u(k) is the control quantity, and L is the linearization length;/> is the optimal estimate of the output at the current moment; μ>0 is the first weight factor, which constrains the change rate of adjacent parameters; 0<η<2 is the first step length factor, which constrains the change rate of adjacent parameters. .

Optionally, the control signal expression is:

Among them, u(k) is the control quantity; is the pseudo partial derivative; u(k) is the control variable; is the pseudo-partial derivative; ρ _i ∈ (0,1] is the second step factor, which is added to make the algorithm more general; v ^* (k+1) is the expected output signal; /> is the optimal estimate of the output at the current moment; λ is the second weight factor.

A model-free adaptive robust control system for high-speed trains, including:

Function acquisition module, used to obtain the improved Kalman filter;

The first data acquisition module is used to obtain the control input of the previous moment and the output optimal estimate of the previous moment based on the high-speed train control system;

An output prediction value determination module, configured to determine the output prediction value at the current moment based on the control input at the previous moment and the output optimal estimate at the previous moment;

The second data acquisition module is used to acquire the output measurement value of the high-speed train at the current moment;

A gain calculation module, used to calculate the gain according to the improved Kalman filter;

A correction module, configured to correct the output prediction value at the current moment according to the gain and the output measurement value at the current moment to obtain the optimal output estimate of the system at the current moment;

A pseudo-partial derivative determination module, configured to determine pseudo-partial derivatives based on the control input at the previous moment, the output optimal estimate at the previous moment, and the output optimal estimate at the current moment;

The third data acquisition module is used to obtain the desired output signal;

A deviation determination module, configured to determine the deviation based on the output optimal estimate at the current moment and the expected output signal;

The control signal determination module is used to determine the high-speed train control signal at the current moment based on the pseudo-partial derivative, deviation and the control input at the previous moment.

An electronic device includes a memory and a processor. The memory is used to store a computer program. The processor runs the computer program to enable the electronic device to execute the model-free adaptive robust control method of high-speed trains. .

A computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the model-free adaptive robust control method of a high-speed train is implemented.

According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

Compared with the existing disturbance suppression MFAC scheme, the present invention has the advantages of good tracking effect, strong applicability and fewer tuning parameters. In addition, the technical solution of the present invention is simple and practical, and can realize the optimization of automatic driving control of high-speed trains.

Description of the drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the drawings needed to be used in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some of the drawings of the present invention. Embodiments, for those of ordinary skill in the art, other drawings can also be obtained based on these drawings without exerting creative efforts.

Figure 1 is a flow chart of the model-free adaptive robust control method of high-speed trains according to the present invention;

Figure 2 is an algorithm structure diagram of the model-free adaptive control scheme for disturbance suppression based on the improved Kalman filter;

Figure 3 is a comparison chart of the velocity-displacement tracking curves of IKF-MFAC and MFAC without considering disturbance;

Figure 4 is a comparison diagram of the velocity-displacement tracking error of IKF-MFAC and MFAC without considering disturbance;

Figure 5 is a comparison chart of the velocity-displacement tracking curves of IKF-MFAC, D-MFAC and MFAC when considering disturbances;

Figure 6 is a comparison chart of the velocity-displacement tracking errors of IKF-MFAC, D-MFAC and MFAC when considering disturbances;

Figure 7 is a comparison chart of the control signal changes of IKF-MFAC, D-MFAC and MFAC when considering disturbance;

Figure 8 is a comparison chart of the acceleration changes of IKF-MFAC, D-MFAC and MFAC when considering disturbance.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

The purpose of the present invention is to provide a model-free adaptive robust control method and system for high-speed trains, thereby achieving suppression of measurement disturbances and improving the response speed of the controller.

In order to make the above objects, features and advantages of the present invention more obvious and understandable, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

The technical solution of the present invention is to use a dynamic linearized data model to replace the train's mechanism model and design an improved Kalman filter (IKF). Based on the improved Kalman filter (IKF), the IKF is combined with the MFAC method to achieve control of systems with unknown mathematical models and measurement disturbances. A disturbance suppression MFAC scheme based on improved Kalman filter is proposed to achieve safe, punctual and energy-saving operation of trains.

Specifically, Figure 1 is a flow chart of a model-free adaptive robust control method for high-speed trains of the present invention. As shown in Figure 1, the method includes:

Step 1: Obtain a modified Kalman filter.

Among them, the steps of the improved Kalman filter (IKF) design method are:

(1) A dynamic linearized data model is used to replace the state equation to describe the dynamic characteristics of the system. A general nonlinear discrete-time system is described by the following dynamic linearized data model:

v _m (k)=v(k)+f(k)

Among them, w(k)∈R is the data process noise, and f(k)∈R is the measurement noise.

(2) Make assumptions about the system: w(k) and f(k) are uncorrelated white noise with mean 0 and variance X and R, as follows:

E[w(k)]＝0,E[w(k) ² ]＝X,E[f(k)]＝0,

E[f(k) ² ]=R,E[w(k)f(k)]=0

Among them, E is the sign of the variance.

The dynamic linearized data model is used to make one-step forward prediction of the real output data. The parameter uncertainty caused by the dynamic linearized data model error and other factors is expressed by the process noise w(k); it is assumed that the noise is the same as the classical Kalman filter. The assumptions of the properties are consistent. For noise complexes with unknown probability distribution and complex sources in the actual environment, Gaussian white noise is a reasonable approximation;

remember For the linear minimum variance estimate of the system output v(k) on the measurement data v _m (k), the following lemma is given:

Linear minimum variance estimator has the following properties:

If the linear minimum variance estimate of the estimated a on the measurement vector z is E*[a/z], then the linear minimum variance estimate of F _x +e on z is:

E*[(Fx+e)/z]＝FE*[x/z]+e

Among them, F is the deterministic matrix and e is the deterministic vector.

(3) Initial output estimate of the given system And after the initial error variance P(0|0)=P(0), when the general nonlinear discrete-time system described by the dynamic linearized data model satisfies the above assumptions, the improved Kalman filter scheme is:

One-step prediction of output data:

One-step prediction error variance: P(k|k-1)=P(k-1|k-1)+X

Calculate the IKF gain:

Output data optimal estimate:

Update estimated error variance: P(k|k)=[1-K(k)]P(k|k-1)

in, Output the best estimate for the previous moment/> one-step prediction,/> Is the best estimate of the output at the current moment;/> is the pseudo partial derivative; U _L (k)=[u(k),...,u(k-L+1)] ^T , u(k) is the control variable, L is the linearization length; P(k-1| k-1) is the estimated error variance at the previous moment, P(k|k-1) is the one-step prediction error variance, P(k|k) is the estimated error variance at the current moment; K(k) is the IKF gain; v _m ( k)=v(k)+f(k) is the output measurement value of the high-speed train system that is disturbed, v(k) is the actual output of the system, f(k) is the measured interference; X is the noise variance.

After the above-mentioned improved Kalman filter design is completed, the subsequent step is the implementation process of the disturbance suppression model-free adaptive control method based on the improved Kalman filter. Please refer to Figure 2 for its algorithm structure.

Step 2: Obtain the control input of the previous moment and the optimal estimate of the output of the previous moment based on the high-speed train control system.

Step 3: Determine the output prediction value at the current moment based on the control input at the previous moment and the output optimal estimate at the previous moment.

Step 4: Obtain the output measurement value of the high-speed train at the current moment.

Step 5: Calculate the gain based on the modified Kalman filter.

Step 6: Correct the output prediction value at the current time according to the gain and the output measurement value at the current time to obtain the optimal output estimate of the system at the current time.

Specifically, steps 2-6 are to use the filtering capability of IKF to obtain the output prediction value at the current moment based on the system input data at the previous moment and the output optimal estimate prediction. Then calculate the IKF gain, and use the gain and the output measurement value v _m (k) at the current moment to pair/> Make corrections to obtain the optimal output estimate of the system at the current moment. The specific calculation process is as follows:

P(k|k-1)＝P(k-1|k-1)+X

in, Output the best estimate for the previous moment/> one-step prediction,/> Is the best estimate of the output at the current moment;/> is the pseudo partial derivative; U _L (k)=[u(k),...,u(k-L+1)] ^T , u(k) is the control variable, L is the linearization length; P(k-1| k-1) is the estimated error variance at the previous moment, P(k|k-1) is the one-step prediction error variance, P(k|k) is the estimated error variance at the current moment; K(k) is the IKF gain; v _m ( k)=v(k)+f(k) is the output measurement value of the high-speed train system that is disturbed, v(k) is the actual output of the system, f(k) is the measured interference; X is the noise variance.

Step 7: Determine pseudo-partial derivatives based on the control input at the previous moment, the output optimal estimate at the previous moment, and the output optimal estimate at the current moment.

Specifically, the following steps are used to determine the estimated value of the pseudo-partial derivative:

Introducing the parameter estimation criterion function:

Derivating the above equation with respect to Φ _P,L (k) and setting it to zero, the following pseudo-partial derivative estimation algorithm is obtained:

In order to enhance the robustness of the parameter estimation algorithm, a parameter reset algorithm is given:

like or |ΔU _L (k-1)|≤ε or/>

but

in, is the pseudo partial derivative; U _L (k)=[u(k),...,u(k-L+1)] ^T , u(k) is the control quantity, and L is the linearization length;/> is the optimal estimate of the output at the current moment; μ>0 is the first weight factor, which constrains the change rate of adjacent parameters; 0<η<2 is the first step length factor, which constrains the change rate of adjacent parameters. .

Step 8: Obtain the desired output signal.

Step 9: Determine the deviation based on the output optimal estimate at the current moment and the expected output signal.

Specifically, the expected output signal v*(k+1) and the optimal output estimate at the current moment are given Make a difference to obtain the deviation e(k) and input it into the MFAC controller.

Step 10: Determine the high-speed train control signal at the current moment based on the pseudo-partial derivative, deviation and the control input at the previous moment.

Among them, the control signal expression is:

Among them, u(k) is the control signal; is the pseudo partial derivative; u(k) is the control variable; is the pseudo-partial derivative; ρ _i ∈ (0,1] is the second step factor, which is added to make the algorithm more general; v ^* (k+1) is the expected output signal; /> is the optimal estimate of the output at the current moment; λ is the second weight factor.

The above control signal is used as an input signal to the high-speed train controlled system to obtain new input and output data. Repeat the above steps to achieve model-free adaptive robust control of the high-speed train.

To sum up, the IKF-MFAC scheme calculates the filter gain at each moment based on the dynamic linearized data model and the statistical characteristics of the measurement disturbance, and calculates the output measurement value v _m (k) measured by the sensor and the model prediction value. Do a reasonable weighted average and calculate the filtered output value/> Then input the MFAC controller for system control. This scheme avoids the problem that the conventional MFAC method directly uses the data output measurement value v _m (k) for controller design, which is easily affected by measurement disturbances, and because/> Derived from data model predictions, there is no need to add additional data measurement equipment.

According to the above solution, the present invention provides an embodiment, specifically as follows:

The implementation of this invention selects the CRH380A EMU running between Jinan and Xuzhou East as the experimental object for simulation research. The maximum output of the train traction unit is 500kN, the maximum output of the braking unit is 500kN, the maximum coupling force of the workshop is 1000kN, and the maximum allowable value of train traction/braking force variation is 60kN/s. The actual speed of the train carriage can be obtained through sensors. The specific parameters of the simulation system are shown in Table 1.

Table 1 EMU model parameters

In order to verify the effectiveness of the IKF-based disturbance suppression MFAC scheme proposed in this invention, simulation comparison experiments of the IKF-based MFAC scheme, the conventional MFAC method and the MFAC scheme with decreasing factor (MFAC with decreasing gain, D-MFAC) are given. Comparative analysis of indicators such as speed tracking effect, control force changes, acceleration changes, etc. was conducted to verify the advantages of the method of the present invention.

Simulation 1: No external disturbance.

The sampling period is set to 1 second, and the sampling samples are 4750. When comparing various control methods, the same model is used to generate data and the initial values are the same.

Figures 3 and 4 show the tracking status and tracking error comparison diagrams of IKF-MFAC and MFAC respectively. When there is no external disturbance, the prediction value correction effect of IKF-MFAC is not obvious, so the tracking effect of the IKF-MFAC method is slightly better than the MFAC method. Both methods can meet the performance requirements in this case.

Simulation 2: There is external disturbance.

When comparing each control method, the same model was used to generate the data, with the same initial values.

The corresponding parameter settings for each control method (IKF-MFAC, D-MFAC and MFAC) are: ε=0.5, v(1)=v*(1), ρ ₁ =ρ ₂ =0.9, eta =μ=1, P(0)=K(0)=0, _em =0.25.

First, the MFAC scheme with attenuation factor is given, as follows:

In the formula, tracking error e(k)=v ^* (k+1)-v _m (k), _em is the set error threshold, and k _m is the switching moment of tracking error e(k). The MFAC scheme with attenuation factor presets an error threshold, and executes the conventional MFAC scheme when the tracking error is large, so that the tracking error first converges to e(k)< _em , and then the attenuation factor comes into play. Since the measurement disturbance is introduced through the error e(k), the measurement disturbance will be suppressed as k increases.

1. Comparison of speed tracking and error between the method of the present invention and other methods:

Figures 5 and 6 show the tracking conditions and tracking errors of the three methods IKF-MFAC, MFAC, and D-MFAC respectively. Due to the suppression effect on disturbances, it is not difficult to see that IKF-MFAC and D-MFAC are better than the prototype MFAC method. The tracking effect is good and the anti-interference ability is obvious. In addition, it can be seen from the figure that since the expected curve is actual line information and the set speed changes frequently, the control effect of the D-MFAC scheme decreases significantly when the set speed changes. The reason is that the time-varying reference signal prevents the system output from entering a steady state, the attenuation factor cannot continue to work, and the selection of the error threshold _em becomes extremely difficult. A small _em cannot suppress the measurement disturbance, while a large _em will make the system output appear step-like and lose smoothness. In this case, the IKF-MFAC scheme that considers filtering the disturbance signal has more advantages.

2. Comparison of control force and acceleration changes between the method of the present invention and other methods:

Figures 7 and 8 are control force and acceleration change diagrams of each control scheme. It can be seen that the unit control force given by each power unit of the IKF-MFAC scheme proposed by the present invention satisfies the constant traction force startup during starting, braking and inertia. , constant power operation and other requirements. In the transition stage of working conditions, the control force can also moderate the changes at a certain rate. The prototype MFAC and D-MFAC have large control force changes when starting, braking, and passing through excessive phases, which brings a certain degree of safety issues to train operation. In addition, the acceleration of the prototype MFAC and D-MFAC changes too frequently and in a large range, which does not meet the comfort requirements of passengers. The high-speed train acceleration transition changes using the IKF-MFAC control method are gentle.

In order to intuitively analyze the control performance of each controller, the following performance indicators are considered to evaluate the controller:

(1) Root mean square error performance (MSE):

Among them, K is the total sampling time.

(2) Maximum acceleration/deceleration (MAXA):

(3) Signal-to-noise ratio (SNR):

MSE calculates the difference between the expected output and the actual output of the system at each moment. The smaller the value, the better the system tracking effect. MAXA calculates the maximum deceleration value during the train tracking process. The larger the value, the more unstable the system operation is. SNR represents the useful signal and noise. The power spectrum ratio of , the larger the value, the better the overall effect of system denoising and control.

Table 2 shows the control performance of the three control methods. From the MSE index, it can be seen that IKF-MFAC has better tracking effect than D-MFAC and prototype MFAC. It can be seen from the maximum acceleration that the maximum acceleration of trains using D-MFAC and prototype MFAC control schemes is relatively large; it can be seen from the SNR index that the IKF-MFAC scheme provided by the present invention is compared with the other two existing schemes. , it can more effectively suppress data measurement disturbances in complex actual lines. The algorithm has better applicability and has smaller tracking root mean square error and greater data signal-to-noise ratio. It can significantly improve the performance of the MFAC method in data measurement disturbances. control performance in the environment.

Table 2 Several performance indicators of each control method

Based on the above method, the present invention also discloses a model-free adaptive robust control system for high-speed trains, including:

Function acquisition module, used to obtain the improved Kalman filter.

The first data acquisition module is used to obtain the control input of the previous moment and the output optimal estimate of the previous moment based on the high-speed train control system.

The output prediction value determination module is configured to determine the output prediction value at the current moment based on the control input at the previous moment and the output optimal estimate at the previous moment.

The second data acquisition module is used to acquire the output measurement value of the high-speed train at the current moment.

A gain calculation module is used to calculate the gain according to the improved Kalman filter.

A correction module, configured to correct the output prediction value at the current time according to the gain and the output measurement value at the current time to obtain the optimal output estimate of the system at the current time.

The pseudo-partial derivative determination module is used to determine the pseudo-partial derivative according to the control input at the previous moment, the output optimal estimate at the previous moment, and the output optimal estimate at the current moment.

The third data acquisition module is used to acquire the desired output signal.

A deviation determination module, configured to determine the deviation according to the output optimal estimate at the current moment and the expected output signal.

The control signal determination module is used to determine the high-speed train control signal at the current moment based on the pseudo-partial derivative, deviation and the control input at the previous moment.

The invention also discloses the following technical effects:

This invention draws on the design ideas of the Kalman filter, derives the improved Kalman filter, and uses a dynamic linearized data model to replace the state equation for design, which can realize the output data containing noise when the controlled system model is unknown. filtering.

Based on the improved Kalman filter, the present invention further designs a disturbance suppression MFAC scheme based on the improved Kalman filter to achieve control of a system with unknown models and measurement disturbances.

The present invention is a disturbance suppression MFAC scheme based on an improved Kalman filter. Compared with the existing disturbance suppression MFAC scheme, this scheme has the advantages of good tracking effect, strong applicability, and fewer tuning parameters. Achieve safe and on-time train operation and ensure passenger safety. This technical solution is simple and practical, and can realize the optimization of automatic driving control of high-speed trains.

Each embodiment in this specification is described in a progressive manner. Each embodiment focuses on its differences from other embodiments. The same and similar parts between the various embodiments can be referred to each other. As for the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple. For relevant details, please refer to the description in the method section.

This article uses specific examples to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method and the core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the present invention There will be changes in the specific implementation methods and application scope of the ideas. In summary, the contents of this description should not be construed as limitations of the present invention.

Claims (7)

1. The model-free adaptive robust control method for the high-speed train is characterized by comprising the following steps of:

obtaining an improved Kalman filter; wherein the improved kalman filter is:

output data one-step prediction:

one-step prediction error variance: p (k|k-1) =P (k-1|k-1) +X

Calculating IKF gain:

outputting the optimal estimation value of the data:

updating the estimation error variance: p (k|k) = [1-K (K) ] P (k|k-1)

Wherein,outputting the optimal estimated value for the last moment +.>Is one-step predictive of->Outputting an optimal estimated value for the current moment; />Is a pseudo partial derivative; u (U) _L (k)＝[u(k),…,u(k-L+1)] ^T U (k) is a control signal, L is a linearization length; p (k-1|k-1) is the estimated error variance at the previous time, P (k|k-1) is the one-step predicted error variance, and P (k|k) is the estimated error variance at the current time; k (K) is the IKF gain; v _m (k) =v (k) +f (k) is an output measurement value of disturbance of the high-speed train system, v (k) is actually output by the system, and f (k) is measurement disturbance; x is the noise variance;

acquiring a control input of the last moment and an output optimal estimated value of the last moment based on a high-speed train control system;

determining an output predicted value of the current moment according to the control input of the last moment and the output optimal estimated value of the last moment;

obtaining an output measured value of the current moment of the high-speed train;

calculating a gain from the modified kalman filter;

correcting the output predicted value of the current moment according to the gain and the output measured value of the current moment to obtain an output optimal estimated value of the current moment of the system;

determining a pseudo partial derivative according to the control input of the previous moment, the output optimal estimated value of the previous moment and the output optimal estimated value of the current moment;

acquiring a desired output signal;

determining a deviation according to the output optimal estimated value at the current moment and the expected output signal;

and determining a control signal of the high-speed train at the current moment according to the pseudo partial derivative, the deviation and the control input at the last moment.

2. The model-free adaptive robust control method for high speed trains according to claim 1, characterized in that the modified kalman filter describes the dynamics of the system using a dynamic linearization data model.

3. The model-free adaptive robust control method for a high speed train of claim 1, wherein the pseudo partial derivative is determined using the formula:

if it isOr |DeltaU _L (k-1) ε or ++>

Then

Wherein,is a pseudo partial derivative; u (U) _L (k)＝[u(k),…,u(k-L+1)] ^T U (k) is a control quantity, and L is a linearization length; />Outputting an optimal estimated value for the current moment; mu > 0 is a first weight factor, and is used for restraining the change rate of adjacent parameters; 0 < eta < 2 is a first step factor, and is used for restraining the change rate of adjacent parameters.

4. The model-free adaptive robust control method of a high speed train of claim 1, wherein the control signal is:

wherein u (k) is a control signal;is a pseudo partial derivative; ρ _i ∈(0，1]Is a second step size factor; v (k+1) is the desired output signal; />Outputting an optimal estimated value for the current moment; λ is a second weight factor.

5. A model-free adaptive robust control system for a high speed train, comprising:

the function acquisition module is used for acquiring an improved Kalman filter; wherein the improved kalman filter is:

output data one-step prediction:

one-step prediction error variance: p (k|k-1) =P (k-1|k-1) +X

Calculating IKF gain:

outputting the optimal estimation value of the data:

updating the estimation error variance: p (k|k) = [1-K (K) ] P (k|k-1)

Wherein,outputting the optimal estimated value for the last moment +.>Is one-step predictive of->Outputting an optimal estimated value for the current moment; />Is a pseudo partial derivative; u (U) _L (k)＝[u(k),…,u(k-L+1)] ^T U (k) is a control signal, L is a linearization length; p (k-1|k-1) is the estimated error variance at the previous time, P (k|k-1) is the one-step predicted error variance, and P (k|k) is the estimated error variance at the current time; k (K) is the IKF gain; v _m (k) =v (k) +f (k) is an output measurement value of disturbance of the high-speed train system, v (k) is actually output by the system, and f (k) is measurement disturbance; x is the noise variance;

the first data acquisition module is used for acquiring a control input at the last moment and an output optimal estimated value at the last moment based on the high-speed train control system;

the output predicted value determining module is used for determining an output predicted value at the current moment according to the control input at the last moment and the output optimal estimated value at the last moment;

the second data acquisition module is used for acquiring an output measured value of the current moment of the high-speed train;

a gain calculation module for calculating a gain from the modified kalman filter;

the correction module is used for correcting the output predicted value of the current moment according to the gain and the output measured value of the current moment to obtain an output optimal estimated value of the current moment of the system;

the pseudo partial derivative determining module is used for determining a pseudo partial derivative according to the control input of the last moment, the output optimal estimated value of the last moment and the output optimal estimated value of the current moment;

a third data acquisition module for acquiring a desired output signal;

the deviation determining module is used for determining deviation according to the output optimal estimated value at the current moment and the expected output signal;

and the control signal determining module is used for determining a control signal of the high-speed train at the current moment according to the pseudo partial derivative, the deviation and the control input at the last moment.

6. An electronic device comprising a memory for storing a computer program and a processor that runs the computer program to cause the electronic device to perform the model-free adaptive robust control method of a high speed train as claimed in any one of claims 1-4.

7. A computer readable storage medium, characterized in that it stores a computer program which, when executed by a processor, implements the model-free adaptive robust control method of a high speed train according to any of claims 1-4.